Parametric sensitivity analysis for models of reaction networks within interacting compartments

Area of Science:

• Computational biology
• Systems biology
• Biophysics

Background:

• Stochastic reaction networks are fundamental in modeling biochemical processes.
• Models of reaction networks in interacting compartments (RNIC) extend these by incorporating dynamic compartment behavior (e.g., cell splitting/merging).
• Efficient parameter sensitivity analysis is vital for understanding complex biological models.

Purpose of the Study:

• To develop and evaluate computational methods for parametric sensitivity analysis in RNIC models.
• To adapt established methods from stochastic reaction networks to the RNIC framework.
• To provide practical tools for researchers studying dynamic biological systems.

Main Methods:

• Implementation of the Girsanov transformation (likelihood ratio method) for unbiased sensitivity estimation.
• Application of coupling methods for finite difference approximations.
• Numerical comparison of these methods within the RNIC setting.

Main Results:

• The Girsanov transformation and coupling methods are effective for sensitivity analysis in RNIC models.
• The relative performance of these methods aligns with their behavior in standard stochastic reaction networks.
• MATLAB codes for all implemented methods are made publicly available.

Conclusions:

• The developed methods provide efficient tools for parameter sensitivity analysis in RNIC models.
• These methods facilitate a deeper understanding of complex biological systems with dynamic compartments.
• Open-source code availability promotes reproducibility and further research in the field.